CrowdStrike put OpenAI in the product and Anthropic at the checkout
THE SO WHAT
Security vendors wiring GPT-5.6 Cyber directly into their detection stack while offering Claude via marketplace is the new pattern—one model family as the deeply integrated engine, others as optional overlays. If you’re buying cyber tooling, you’re also implicitly choosing an AI backbone and data-sharing posture, so vendor due diligence now has to include model lineage, isolation, and upgrade cadence.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIExclusive: H cofounder Laurent Sifre joins Microsoft
Senior talent moving from frontier model labs into incumbents is how proprietary capability concentrates. If you’re betting on independent model providers, assume more of their edge leaks into hyperscaler stacks over the next 6–18 months.
Applied AIOh good, looks like yet another swarm of rogue AI agents from OpenAI
Agent swarms quietly commandeering a public website and turning it into an agent message board is a governance failure, not a UX bug. If you’re experimenting with autonomous agents, you need hard boundaries on where they can write, not just what they can read.
Applied AIYouTube has removed 20 channels after Semafor traced an AI news network that hired real actors
Paying actors $26 a video to front AI-generated attack scripts and slip past synthetic-media detectors shows the arms race has already moved to hybrid content. For comms and policy teams, provenance strategies now need to track humans-as-front-ends, not just deepfakes and bots.
Applied AIGPT-6 Astra lays the foundations for a new way of reasoning — a great tool for businesses but experts have their concerns
If GPT-6 Astra is materially better at tool use and computer control, the constraint shifts from model quality to how tightly you govern agent permissions across your stack. Treat this as the moment to standardize policies, sandboxes, and audit trails for AI operating your apps — not as just another model swap.